Rally Apr. 17th-19th

[gd_barnes]

Originally Posted by IronBits

They can if it's one of them pants/skirt things.

I'm bringing quite a few cores, how many are the rest of you bringing to the Rally?
Are you going to let Free-DC slam dunk you again ????

40 fast and 2 slower here. I'll PM Lennart shortly to give ROLP a fighting chance.

[Mini-Geek]

Originally Posted by gd_barnes

Point taken on the 1st part. Although I never claimed my posts were more interesting than the math, perhaps I implied it. (lol)

I don't know if you implied it, just pointing out that I don't think my posts are.

Originally Posted by gd_barnes

On the 2nd one, [stuff]

Sounds a bit tricky, but I guess you could get used to it. Thanks for the explanation, it makes sense now.

Originally Posted by IronBits

I'm bringing quite a few cores, how many are the rest of you bringing to the Rally?

Just my two, as usual. (big surprise there, huh?)

[Flatlander]

There is a skip just down the road but there were no PCs in it, so I'll be bringing my usaul 6 cores.

Where does the constant "1.781" come from in C10 of the spreadsheet?

Re. pants:
http://www.youtube.com/watch?v=S8pJIiV9KWo

[mdettweiler]

@Gary regarding your big post about probability and odds: ahh, I think I get it now. So, I guess one could say that, on average, if for a given range of numbers the odds of a prime are 1 in 4000, then someone will most likely score at least one prime (or, to be more exact, 1.15 primes ), if they do ~4600 tests--correct?

Oh, one last thing that I just thought of: is there any difference between the terms "1 in x odds" and "a chance of x to 1", besides the obvious reciprocation of the placements of x and 1? If there is a difference, then maybe that was at least a partial contributor to my confusion, since I'd always thought they were equivalent.

(P.S.: I'll play around with the spreadsheet a bit later today as you suggest; hopefully that will shed further light on this topic. )

[gd_barnes]

Originally Posted by mdettweiler

@Gary regarding your big post about probability and odds: ahh, I think I get it now. So, I guess one could say that, on average, if for a given range of numbers the odds of a prime are 1 in 4000, then someone will most likely score at least one prime (or, to be more exact, 1.15 primes ), if they do ~4600 tests--correct?

Oh, one last thing that I just thought of: is there any difference between the terms "1 in x odds" and "a chance of x to 1", besides the obvious reciprocation of the placements of x and 1? If there is a difference, then maybe that was at least a partial contributor to my confusion, since I'd always thought they were equivalent.

(P.S.: I'll play around with the spreadsheet a bit later today as you suggest; hopefully that will shed further light on this topic. )

Yes, correct on your first paragraph! Cool, I'm glad you got it. It's very strange but true. Keep in mind that they still will "most likely" score one prime at 4000 tests. They have a 62% chance of doing so. It's just that 38% of the time they won't find a prime. If that was all there was, then it would only amount to an average of 0.62 primes per 4000 tests, which is obviously incorrect. What brings the average up to 1 prime in 4000 tests is when 2 or more primes are found during that span.

BTW, you could check this in all of your results files. If you have enough 6500-test samples around n=~500K (odds are 1 in ~6500 of prime), you'll come in close to ~62% of them will contain a prime, although some will contain 2 or more.

On your 2nd paragraph, you came to the right guy for that. :-) In roulette, they pay you 35 to 1 if your number hits but you get to keep the bet that you placed. Therefore it also can be stated 36 FOR 1. Some casinos will try to "mess" with you by quoting odds different ways like that. So in order to break even at roulette over the long run, your number has to hit 1 out of every 36 times.

Therefore for something that happens 1 out of every 36 times you have:
Odds of 1 in 36 -or-
Chances of 35 TO 1 -or-
Chances of 36 FOR 1

Of course if you've played roulette, you know that the true odds are 1 in 38 since there are 38 #'s on the (American) roulette wheel. That's the house "vig" of 5.26% there. [For the Europeans around, I realize that it is 1 in 37 where the house vig is only 2.63%. It's a far more popular casino game in Europe than the U.S. The better odds for the player are probably a big part of it.]

So no, you cannot just reciprocate the placement of the odds when stating it in xx to xx format. But also no, I do not believe that was what caused your confusion. I believe that was caused by forgetting that you can actually find 2 or more primes in a 1000 test sample. So even for a 1 in 4000 chance of something, you have to run 1150 tests to have a 25% chance of it occuring but you COULD find 2 or more primes during that time.

Making it easier, sometimes you'll hear someone say something has about an "even chance" of happening or that it is an "even chance" bet. What that means is that the chance of it happening or that the bet pays is 1 TO 1. (The two 1's is why it is referred to as even chance; i.e. they are the same.) If you win, you get paid $1 and you keep your original bet or stated differently 2 FOR 1 or odds of 1 in 2.

If you ever hear about a horse or dog going off at 3 to 5, you'll know he's very good. People are willing to bet $5 to win $3. The animal has to win 5 out of every 8 times for you to break even in the long run. (I've seen one go off at 1 to 5 but it is quite rare. It would require that the animal win 5 out of 6 times for you to break even in the long run.) Almost always the left side of the xx to xx form is higher.

BTW, if I remember right, about a month ago, the Patriots were about a 4 to 1 or 7 to 2 (i.e. 3.5 to 1) favorite to win the Super Bowl next year. That's probably pretty good considering there are 32 teams. I think the odds on various things like that are quite interesting but have never placed a sports bet in my life. I think a week before the presidental election, Obama was about a 1 to 7 favorite to win. Obviously McCain was given little chance a week out.

OK, rambled and gambled again.


Gary

[mdettweiler]

Ah, I get it now. Thanks--that makes sense.

[IronBits]

Start 00:00:00 April 17th GMT right?
Ends 00:00:00 April 20th GMT right?

[Mini-Geek]

Originally Posted by IronBits

Start 00:00:00 April 17th GMT right?
Ends 00:00:00 April 20th GMT right?

Originally Posted by gd_barnes

...from April 17th to 19th starting and ending at the usual 7 PM GMT.

So no. Start 19:00:00 April 17th GMT, Ends 19:00:00 April 19th GMT.

[PCZ]

LA 12am Friday 17th
SA 2pm Friday 17th
NY 3pm Friday 17th
Montreal 4pm Friday 17th
London 8pm Friday 17th
Paris 9pm Friday 17th
Berlin 10pm Friday 17th
Delhi 1.30am Saturday 18th
Sydney 5am Saturday 18th

Funny old thing time it's all over the place :)

[gd_barnes]

We attempt to start and end it a time that is convienient for the most people, i.e. in the U.S. and Europe.

We'll leave it as is for this rally but with Dave (AMDave) in Australia, we could consider starting them a little later, perhaps at 9 PM GMT. That'd still be early (7 AM) in Sydney but not as bad as 5 AM and it would allow a few folks to be home from work in the U.S., especially in the summer. It would only push it to midnight in Berlin so that wouldn't make it too bad for Karsten.

I don't think we have anyone east of Germany except for Dave so that might actually be a better alternative for future rallies.

That said, I'm not sure it makes much difference on the start and end time. I think that people usually just set their machines to crunch before they go to work or bed and let 'er rip. I usually get them all crunching except my slower cores a day or more before the rally.


Gary

[IronBits]

GMT -0 1900
EDT -4 1500
CDT -5 1400
MDT -6 1300
MST -7 1200 (Arizona)
PDT -7 1200